Journal of Power Sources 412 (2019) 451–464

Contents lists available at ScienceDirect

Journal of Power Sources

journal homepage: www.elsevier.com/locate/jpowsour

Modeling of oxygen reduction reaction in porous carbon materials in T

alkaline medium. Effect of microporosity
Atsushi Gabea, Ramiro Ruiz-Rosasa, Carolina González-Gaitánb, Emilia Morallónb,
Diego Cazorla-Amorósa,∗
a
Instituto Universitario de Materiales, Departamento de Química Inorgánica, Universidad de Alicante, Apartado 99, 03080, Alicante, Spain
b
Instituto Universitario de Materiales, Departamento de Química Física, Universidad de Alicante, Apartado 99, 03080, Alicante, Spain

HIGHLIGHTS GRAPHICAL ABSTRACT

• Activity of porous carbons in ORR is

undoubtedly related to their porosity.
• ORR
HO
occurs through O reduction to
−
and then HO −
2
reduction to
2 2
−
OH .
• Microporosity is correlated to a high
activity of both O and HO −
reduc-
2 2
tions.
• Narrow micropores have higher ac-
tivity to HO −
reduction.
2

ARTICLE INFO ABSTRACT

Keywords: The role of porosity, and more specifically, microporosity, in the performance of carbon materials as Oxygen
Oxygen reduction reaction Reduction Reaction (ORR) catalysts in alkaline medium still has to be clarified. For this purpose, a highly
Hydrogen peroxide reduction microporous KOH-activated carbon and a microporous char have been prepared and their ORR performance in
Microporosity alkaline media were compared to that of two commercial carbon blacks with low and high surface areas, re-
ORR mathematical modeling
spectively. Interestingly, all carbon materials show a two-wave electrocatalytic process, where the limiting
Charge transfer reaction
current and the number of electron transferred increase when going to more negative potentials. The limiting
Mass transfer rate
current and onset potential of the second wave is positively related to the amount of microporosity, and H2O2
electrochemical reduction tests have confirmed that the second wave could be related to the catalytic activity
towards this reaction. In accordance to these findings, a model is developed that takes into account narrow and
wide micropores in both charge transfer reactions and the mass transfer rate of O2 and H2O2. This model suc-
cessfully reproduces the experimental electrochemical response during ORR of the analyzed porous carbon
materials and suggests the important role of narrow micropores in H2O2 reduction.

1. Introduction
Due to their low cost, high stability, high electrical conductivity and
tunable surface chemistry and porosity, carbon materials are attractive
materials for their use in catalysts for electrochemical energy
conversion [1]. In the case of low temperature fuel cells, carbon ma-
terials are the main component of the electrodes, where they can be
employed as a support and, more recently, even as catalyst by them-
selves. Nowadays, carbon black currently constitutes the preferred
support in commercial carbon catalyst layers [2,3].

∗
Corresponding author. Universidad de Alicante Carretera de San Vicente del Raspeig, s/n 03690 San Vicente del Raspeig, Alicante, Spain.
E-mail address: cazorla@ua.es (D. Cazorla-Amorós).

https://doi.org/10.1016/j.jpowsour.2018.11.075
Received 23 July 2018; Received in revised form 1 November 2018; Accepted 23 November 2018
Available online 30 November 2018
0378-7753/ © 2018 Elsevier B.V. All rights reserved.
A. Gabe et al. Journal of Power Sources 412 (2019) 451–464

Nomenclature micropores, mol cm−2 s−1

JHD2, wp
O2 Hydrogen peroxide diffusion rate in wide micropores, mol
Cg gravimetric capacitance, F g−1 cm−2 s−1
Cij Concentration of i reagent at j position (bulk, narrow or JOD2, j Oxygen diffusion rate in narrow or wide micropores, mol
wide micropores), mol cm−3 cm−2 s−1
COb2 oxygen concentration in the bulk of the solution, mol JHK2, jO2 intrinsic electrochemical reaction rate for H2O2 reduction
cm−3 in narrow or wide micropores, mol cm−2 s−1
Cinp concentration of i component in narrow micropores of the JOK2, j intrinsic electrochemical reaction rate for O2 reduction in
electrode, mol cm−3 narrow or wide micropores, mol cm−2 s−1
Ciwp concentration of i component in wide micropores of the k 0,r ,ij charge transfer rate constant of i component at narrow or
electrode, mol cm−3 wide pores, cm s−1
CO2TPD CO2-evolving surface oxygen groups from TPD, μmol g−1 kir kinetic constant for the electrochemical reduction of i
COTPD CO-evolving surface oxygen groups from TPD, μmol g−1 component, cm s−1
CXPS carbon atomic concentration by XPS, at.% Kiads adsorption constant for the i component, cm3 mol−1
D O2 diffusion coefficient of oxygen, 1.95·10−5 cm2 s−1 at 25 °C L thickness of the electrodic layer of carbon material, cm
np
DH La crystallite thickness along the a-axis for a graphitic struc-
2O 2
effective mass transfer coefficient of H2O2 within the
2·L ture, nm
narrow micropores, cm s−1
Lc crystallite thickness along the c-axis for a graphitic struc-
D H 2O 2
effective mass transfer coefficient of hydrogen peroxide,
2·L ture, nm
0.0074 cm s−1 at 1600 rpm
n number of transferred electrons
D O2
effective mass transfer coefficient of oxygen,
2·L OTPD total evolved oxygen from TPD, μmol g−1
0.0125 cm s−1 at 1600 rpm
OXPS oxygen atomic concentration by XPS, at.%
E electrode potential vs the reversible hydrogen electrode, V
RH2O2 k 0,r H2 O2 / k Hf 2 O2 ratio
F Faraday constant, 96485 C mol−1
SBET apparent surface area from BET method, m2 g−1
ID current measured at the disk of the rotating ring-disk
VDRCO2 Dubinin-Raduskevich micropore volume from CO2 ad-
electrode, mA
sorption isotherm, cm3 g−1
IR current measured at the ring of the rotating ring-disk
VDRN2 Dubinin-Raduskevich micropore volume from N2 adsorp-
electrode, mA
tion isotherm, cm3 g−1
j specific current measured on the disk of the rotating ring-
Vmes mesopore volume from N2 adsorption isotherm, cm3 g−1
disk electrode, mA cm−2
electron transfer coefficient of i component
modeled specific current for the electrochemical reduction
i
jH2 O2
υ kinematic viscosity of 0.1 M KOH electrolyte, 0.01 cm2 s−1
of H2O2 to H2O, mA cm−2
ω rotation rate, rad s−1
jO2 modeled specific current for the electrochemical reduction
electrode overpotential, V
of O2 to H2O2, mA cm−2
jHL 2O2 Levich limiting current density for hydrogen peroxide, mA ratio between effective surface area and geometric surface
cm−2 area of disk
jOL2 Levich limiting current density for oxygen, mA cm−2 * active site for ORR
jDORR modeled specific current on the disk of the rotating ring- *f free amount of active site
disk electrode, mA cm−2 *t total number of active sites
JHD2, np Hydrogen peroxide internal diffusion rate in narrow *(i) adsorbed ORR intermediate on active site
O2

Much attention is being paid to the development of novel catalysts
for the oxygen reduction reaction (ORR) because it is a slow reaction
that necessitates high Pt loadings. This reaction takes place at the
cathode of all fuel cell systems and other energy storage devices such as
metal-air batteries. The high cost, low chemical stability and perfor-
mance of current platinum-based electrodes are the main obstacles for
the commercial implementation of these systems, therefore explaining
the growing interest in non-noble or metal-free ORR catalysts [4–7].
Even in the case of noble-metal based catalysts, the contribution of
carbon support to ORR is far from being negligible, and can account for
values as high as one third of the power delivered by a fuel cell [8], and
can severely decline the stability of the cathode and the membrane
when the 2-electron pathway towards hydrogen peroxide prevails [9].
This explains the huge interest in determining the parameters that
drives the ORR activity of carbon materials. The ORR activity of un-
doped carbon materials is low. The highest activity is achieved at high
pH values, although the reaction rate still is sluggish, and the reaction
delivers the formation of hydroperoxide ion through a 2-transferred
electron process [10]. Nevertheless, the surface chemistry and the
textural parameters of carbon materials can be modified for enhancing
their electrocatalytic activity.
The effect of the surface chemistry on the ORR activity was profu-
sely studied in recent years [5,11,12]. [13]. [14], and a great attention
was devoted to understand the effect of the surface chemistry on the
catalysis of ORR by carbon materials and to reveal the nature of the
active site [9,11,14–21].
However, the effect of porosity on the kinetics of ORR by carbon
materials is often ignored, and the presence of catalytic species makes
challenging any clear assessment [22–27]. Few works can be found
where a systematic approach for assessing the role of microporosity of
bare carbon materials is reported. In this sense, Appleby and Marie
reported ORR kinetic studies carried out on a large collection of carbon
materials in alkaline solution [28]. They found that ORR activity in-
creased linearly with BET surface area for carbon blacks, whereas no
clear trend was found in the case of activated carbons (ACs). They ar-
gued that, although the high surface area of ACs should provide a larger
amount of active sites for ORR, those lying on micropores are un-
reachable either for the electrolyte (unwetted pores), for dissolved
oxygen or for the solvated HO2− anion, explaining why their catalytic
activity is lower than expected considering their BET surface areas.
Recently, Liu et al. analyzed the influence of micro and mesoporosity on
ORR cathode catalysts [29], and they concluded that activity increases
by the presence of microporosity and that mesoporosity is necessary to
facilitate accessibility to active sites. On the other hand, Seredych et al.
prepared ORR catalysts using a hydrophobic ultramicroporous carbon
made from furfuryl alcohol and tannin [30]. The ORR experiments in
rotating disk electrode revealed an inverse relationship between the
micropore size and volume (especially those with sizes lower than

452
A. Gabe et al.
0.7 nm) and the catalytic activity at low potentials. The authors pro-
posed that a strong adsorption of oxygen takes place in hydrophobic
ultramicropores, leading to weakening of OeO bonds and promoting
the dissociation of dioxygen. Again, the presence of a wider porosity of
hydrophilic character is considered critical for achieving a good oxygen
mass transfer rate to the ultramicropores. Finally, Qu reported the ORR
catalytic activity of several ACs of different micropore sizes [31].
However, most of the work is devoted to find a relationship between the
activated carbon structure and crystallinity rather than in the effect of
microporosity, which is considered to be “not in use” in the gas diffu-
sion layers of catalysts.
This work studies the effect of porosity, and in particular micro-
porosity, in carbon material electrodes in the ORR in alkaline condi-
tions. For this purpose, carbon materials with different textural prop-
erties but similar surface chemistry were selected. This collection
includes a highly microporous activated carbon, a char with a low
microporosity development and two commercial carbon blacks of dif-
ferent surface area and pore size distributions. After determining the
ORR activity of these materials, a kinetic model for the oxygen reduc-
tion reaction over porous carbon materials that takes into account the
critical role of microporosity in the ORR is drawn. Using this model, the
current density of porous carbon materials in ORR is derived and its
validity for describing the experimental results of the porous carbon
samples is discussed.
2. Experimental section
2.1. Material synthesis
A highly microporous activated carbon (KUA) was prepared by
chemical activation of Spanish anthracite with KOH. The impregnation
ratio of KOH to carbon precursor was set to 4:1. The activation was
carried out at 750 °C (heating rate: 5 °C/min, holding time: 2 h) in a
tubular furnace equipped with a quartz tube feed with 800 mL/min of
nitrogen (99.999%, Air Liquide). The pyrolyzed mixture was stirred in
5 M HCl and distilled water until neutral pH. The activated carbon was
recovered by filtration and dried at 120 °C. More details about the
preparation are found elsewhere [32]. Microporous char (AC) was
prepared by carbonization of a phenolformaldehyde polymer resin in
N2, at 5 °C/min to 1273 °C with 1 h soaking time. Commercial Vulcan
XC-72F carbon black (XC72) from Cabot corporation and CD-6008
carbon black (CB) supplied from COLUMBIAN CHEMICALS were used
as received.
Fig. 1. A) N2 adsorption-desorption isotherms at −196 °C and B) derived NL-DFT pore size distributions of KUA, AC, CB, and XC72.
453
Journal of Power Sources 412 (2019) 451–464
2.2. Material characterization
The porosity of the carbon materials was analyzed by CO2 and N2
adsorption isotherms at 0 °C and −196 °C, respectively, using an
Autosorb-6B apparatus (Quantachrome). The structure of the samples
was characterized by transmission electron microscopy (TEM) and X-
ray Diffraction (XRD). The surface chemistry was assessed using X-Ray
Photoelectron Spectroscopy (XPS) and Temperature Programmed
Desorption (TPD) experiments. More details are in the electronic sup-
plementary information (ESI).
2.3. ORR activity experiments
Electrochemical activity tests towards ORR were conducted at 25 °C
in a thermostated three-electrode cell filled with 0.1 M KOH solution
using an Autolab 302N potentiostat (Metrohm). Further explanation is
found in ESI. The number of electrons transferred in the ORR (n) was
followed during the LSV measurements from the oxidation of hydrogen
peroxide over the platinum ring disk using the following equation:
4·ID
n=
ID + IR /N (1)
Where IR and ID stand for the currents measured at the ring and the disk,
respectively, and N is the collection efficiency of the ring, which was
experimentally determined to be 0.37. The number of electrons was
also determined from the slope of the Koutecky -Levich plots at dif-
ferent potentials [33] using LSV recorded at several rotation rates (400,
625, 900, 1225 and 1600 rpm).
The limiting current during RDE experiments ( jOL2 , mA cm−2) was
defined in accordance to Levich model for RDE [33]:
jOL2 = n ·F ·0.62·DO2/3
2
· 1/6 · 1/2 · C B
O2 = n ·F ·kLf, O2 ·COB2 (2)
In Eqn. (2), n is the number of electrons transferred, F is the Faraday
constant (96485 C mol−1), DO2 is the diffusion coefficient of oxygen
(1.95·10−5 cm2 s−1), υ is the kinematic viscosity of the solution
(0.01 cm2 s−1) and ω is the rotation rate (rad s−1). The parameter
0.62· DO2/3
2
· 1/6 · 1/2 has been so-called mass transfer coefficient of
oxygen having a value of 0.0125 cm s−1 for 1600 rpm. The oxygen
concentration in the solution, COB2 , has a value of 1.2·10−6 mol cm−3.
Finally, additional LSV experiments in N2e and O2-saturated 0.1 M
KOH solution containing 3 mM H2O2 were performed using the same
experimental conditions reported for the ORR experiments.
A. Gabe et al.
3. Results and discussion
3.1. Characterization of porosity and structure
Fig. 1A compiles the N2 adsorption-desorption isotherms at −196 °C
for the carbon materials. The samples have different porosities, as de-
duced from the shape of the adsorption isotherms. KUA sample shows a
type I isotherm according to IUPAC classification, which is character-
istic of microporous solids [34], as well as the highest N2 uptake. The
rounded knee at low relative pressures (0–0.2) points out the presence
of a wide micropore size distribution. AC shows a type I plus type IV
isotherm shape, with a much lower N2 uptake than KUA. The hysteresis
loop on the desorption branch of the isotherm evidences the presence of
mesopores. CB sample also exhibits a combination of type I and IV
isotherms although with a large mesoporosity development. Finally,
XC72 sample shows a type II isotherm, typical of non-porous solids
where multilayer adsorption takes place in the external surface of the
particles.
Table 1 summarizes the porosity parameters of all the samples. In
the case of XC72 and CB, most of the surface area (SBET in Table 1)
comes from the external surface of the carbon black particles, since the
contribution of microporosity is small when compared to the total pore
volume of the sample (less than 25%). KUA has a well-developed mi-
croporosity (VDRN2 in Table 1) and also presents some amount of me-
sopores (Vmeso in Table 1) [32,35]. AC and CB share similar VDRN2 and
SBET values, but the mesoporosity development is very different, with
Vmeso being 4 times larger in the carbon black sample.
As for the pore volumes, AC contains both micropores and meso-
pores with a larger mesopore contribution than KUA (10% of total pore
volume are mesopores for KUA, and 45% for AC). CB also combines
micropores and wide mesopores, but with an even larger impact of
mesoporosity (more than 60% of total porosity). Finally, XC72 did not
present any relevant microporosity. It is important to note that the
porosity of CB and XC72 samples, is mainly related with the particle
size, with micro- and mesopores resulting from the void spaces between
the nanosized particles (see Ref. [36] for XC72 and TEM images for CB
sample in Fig. S1). We have also evaluated narrow microporosity (pore
sizes lower than about 0.7 nm) using CO2 adsorption at 0 °C [37,38].
The similar micropore volume obtained by N2 and CO2 adsorption in
AC revealed the presence of a narrow micropore size distribution with
sizes around 0.7 nm in this sample, whereas CB and KUA have a larger
contribution of wide micropores (VDRN2 > VDRCO2).
The pore size distributions (PSD) for the materials were calculated
from the N2 adsorption isotherms using NL-DFT method as proposed by
Jagiello and Olivier [39] using SAIEUS© software (Fig. 1B). These
curves confirmed the wide micropore size distributions for KUA and the
narrow micro PSD for the other materials. KUA has most of its porosity
covering all the micropore range; however, its wide pore size dis-
tribution extends into the narrow mesoporosity region, what explains
the origin of the Vmes value reported in Table 1. In the mesopore region,
it is possible to see that AC and the carbon blacks show a broad pore
size distribution, with mesopores of around 10 nm being the most fre-
quent ones for AC and CB, and 25 nm for XC72 (see inset of Fig. 1B).
XRD patterns of all samples were recorded to provide information
about their crystalline structure (Fig. S2). In agreement with the pre-
sence of micropores, a large X-Ray scattering at low angles (below 10°)
is clearly observed in porous carbon samples, being this feature max-
imized in the case of the highly porous KUA. The lattice parameters
were determined from the (002) and (100) peaks [40] and compiled in
Table 1. The data show that the carbon blacks have a larger structural
ordering than the KOH-activated carbon, showing higher sizes of or-
dered domains in the parallel and perpendicular directions (La and Lc in
Table 1), whereas AC has an intermediate ordering.
454
Journal of Power Sources 412 (2019) 451–464
3.2. Characterization of surface chemistry
The surface chemistry was analyzed by XPS and TPD, Table 2. The
microporous activated carbon has the highest oxygen content, while
both carbon blacks show the lower amounts, as can be deduced from
the XPS results (see Table 2). The absence of relevant amounts of other
impurities such as sulfur, boron, iron or other heteroatoms was also
confirmed. TPD experiments were performed to analyze the surface
functional oxygen groups, Fig. S3. Table 2 summarizes the CO, CO2 and
the total amount of oxygen (O) calculated as CO+2(CO2). When the
evolved amounts are normalized in terms of BET surface area, the
amounts of evolved CO2 are lower in KUA (0.10 μmol m−2 for KUA and
ca. 0.5 μmol m−2 for the rest of samples), and those for CO are higher
for AC (values of 0.54, 2.42, 1.00 and 1.05 μmol m−2 are found for
KUA, AC, CB and XC72).
3.3. Electrochemical characterization
The electrochemical behavior of the samples was analyzed by cyclic
voltammetry in N2 saturated 0.1 M KOH between 0 and 1 V at
50 mV s−1 (see discussion in ESI and Fig. S4). Linear sweep voltam-
metry experiments were also recorded in O2-saturated electrolyte at
10 mV s−1, Fig. 2. As potential was decreased from 1.0 V to less positive
values, all samples showed a reduction peak in presence of O2 with a
net reduction current when the CV is compared to that recorded in N2-
saturated electrolyte. Onset potentials (defined as the point where the
slope of the CV cathodic scan in O2-saturated electrolyte starts to de-
viate from that recorded in N2-saturated electrolyte) were 0.81 V for
XC72 and AC, 0.84 V for KUA and 0.83 V for CB. Further experiments
using a RRDE were performed.
3.4. ORR rotating ring disk measurements
Fig. 3 shows the LSV of all samples at rotation rate of 1600 rpm. The
electrodes were stabilized by cycling for 10 min in O2-saturated KOH at
10 mV s−1. After that, the background current was recorded in N2-sa-
turated electrolyte. The LSV results reported were collected after these
steps, with the ORR current being corrected by subtraction of the
double layer current. The limiting currents for oxygen reduction
through 2 and 4 e− reaction pathways obtained from the Levich theory
(2.9 and 5.8 mA cm−2 from eqn. (2)) at the selected experimental
conditions are also plotted (dashed grey horizontal lines in Fig. 3A).
The onset potential was determined at j = −0.1 mA cm−2 on the LSV
curves in Fig. 3A. XC72 has the lowest ORR activity (onset potential of
0.76 V). AC and CB show similar onset potential, 0.80 and 0.79 V, re-
spectively. Remarkably, the activated carbon has a higher onset po-
tential being 0.85 V. Two-wave shape on their LSVs can be observed
being less pronounced in the case of KUA. These voltammetric profiles
are observed independently of the selected rotation rate, Fig. S5, what
demonstrates that this behavior is not related to external mass transfer
issues. The same study was performed using the bare surface of the
glassy carbon disk, Figs. S5E and F, showing that the activity of the disk
surface is much lower than that for the tested samples, and it always
shows a two-wave shape. Moreover, the diffusion controlled region on
the glassy carbon surface starts at potentials lower than 0.3 V, con-
firming that LSV measurements correspond to the deposited carbon
Table 1
Textural and structural parameters of the analyzed samples.
Sample SBET m2 VDRN2 cm3 VDRCO2 cm3 Vmeso cm3 Lc nm La nm
g−1 g−1 g−1 g−1
KUA 3300 1.23 0.72 0.15 0.6 2.6
AC 590 0.24 0.26 0.22 0.9 3.6
CB 580 0.26 0.17 0.88 1.2 3.7
XC72 280 0.11 0.05 0.33 1.5 4.0
A. Gabe et al. Journal of Power Sources 412 (2019) 451–464

Table 2 2-electron pathway. An increase on the polarization of the electrode

Surface chemistry of the analyzed samples. shifts the ORR to a more favorable pathway involving a higher number
Sample CXPS at.% OXPS at.% COTPD μmol CO2TPD μmol OTPD wt.% of electrons (2.4, 2.7, 2.6 and 3.0 at 0.05 V for XC72, CB, AC and KUA
g−1 g−1 samples according to RRDE experiments, and 2.5, 3.1, 2.8 and 3.4 ac-
cording to K-L theory). The increase in the number of transferred
KUA 90.7 8.8 1780 330 3.9
electrons at that potential region seems to be correlated with the pre-
AC 94.2 5.8 1430 340 3.5
CB 97.7 2.3 580 280 1.8
sence of microporosity, Table 1, rather than with any other textural,
XC72 98.0 2.0 230 120 0.6 structural or surface chemistry parameter. This would be in agreement
with recent findings where ORR activity of milled N-doped carbon
nanotubes was reported to increase due to the formation of micro-
porosity [22].

3.5. Mass loading effect in ORR

Measurements were conducted using two additional sample load-

Fig. 2. Linear Sweep Voltammetry profiles in N2e (dashed lines) and O2e
(solid lines) saturated 0.1 M KOH solutions. KUA (blue), AC (red), CB (black)
and XC72 (green). Scan rate: 10 mV s−1. Electrode loading: 0.4 mg cm−2.
materials.
The number of transferred electrons during ORR was calculated
from the ring to disk current ratio at 1600 rpm and from Koutecky-
Levich (K-L) in eqn. (1), Fig. 3B. ORR activity in these samples involve a
ings (0.10 and 1.00 mg cm−2) for KUA, Figs. S6A and B. LSV curves
were also expressed in gravimetric terms, Figs. S6C and D. Both kinetic
and diffusion rates are negatively impacted by a low surface loading, as
manifested by the current decrease and the onset potential being shifted
to less positive potentials at 0.10 mg cm−2. In addition, the shape of the
LSV profile becomes similar to that of bare glassy carbon surface at
medium and low potentials (see dotted, blue line vs dotted, black line in
Fig. S6A). Decline on ORR activity from very low catalyst loading in
RRDE experiments was reported in the past, and it was accounted to the
catalyst particles being unable to fully cover the whole surface of the
disk, rendering a lower effective surface of the electrode [41]. It was
also proposed that a high diffusion film resistance in the electrode could
result from excessive thickness, reducing the current density [42]. The
effect of the different loadings is easily observed when the LSV profiles
are normalized in terms of weight. It can be seen that the normalized
current near the onset potential goes through a maximum at medium
surface loadings, Fig. S6C, confirming that kinetic control is reached at
that loading, while lower or higher loadings have a negative impact on
the ORR activity due to the aforementioned reasons.
Differently, an improvement in the mass transfer rate on the diffu-
sional controlled region (i.e. higher limiting current due to higher
oxygen diffusion rate) is observed when raising the loading from 0.40 to
1.00 mg cm−2, Fig. S6A. The impact of the ratio between the effective

Fig. 3. Oxygen reduction reaction study from the RRDE at 1600 rpm in 0.1 M KOH solution. A) LSV in the disk electrode (scan rate: 5 mV s−1, capacitive currents
corrected). B) number of electrons measured from the ring vs disk currents (full lines) and from Koutecky-Levich theory (dots). KUA (blue), AC (red), CB (black) and
XC72 (green). Catalyst loading: 0.4 mg cm−2. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this
article.)

455
A. Gabe et al.
surface area available for oxygen diffusion and reduction and the
geometrical area of the disk (which is the used one in the Levich theory
for determining oxygen diffusion rate on a rotating flat electrode sur-
face) in the oxygen mass transfer rate on RDE measurements was al-
ready exposed in the literature [33,43,44]. At 0.40 mg cm−2, the disk
surface is not fully covered by KUA particles. When KUA loading in-
creases, so does the effective area of the electrode. Therefore, the higher
current registered in the diffusional controlled region for
1.00 mg cm−2, Fig. S6A, is in agreement with recent findings regarding
an enhanced ORR activity in RDE experiments of porous electrodes and
nanoparticles-based catalysts due to a higher effective surface area
[43,45,46]. As expected, the gravimetric current values in the diffu-
sional controlled region decrease with the loading of the catalyst, Fig.
S6D. In this region, all oxygen molecules are electrochemically reduced
once they reach the electrodic surface, and therefore the current is only
determined by the oxygen diffusional flow towards the electrode and
the number of transferred electrons during the reaction (i.e. limiting
current is reached).
The number of electrons transferred is positively related to the
surface loading, Fig. S6B. We propose that n values between 2 and 4 e−
observed once the diffusion-controlled region is reached, are due to the
Fig. 4. A) LSV profiles of hydrogen peroxide reduction ([H2O2]: 3 mM) obtained using RRDE at 1600 rpm in N2-saturated 0.1 M KOH. B) KUA, C) AC and D) CB LSV
at 1600 rpm in 0.1 M KOH in H2O2 and N2-saturated electrolyte (dotted lines), H2O2 and O2-saturated electrolyte (solid lines), and O2-saturated electrolyte (dashed
lines).
456
Journal of Power Sources 412 (2019) 451–464
occurrence of two electrochemical reduction reactions: i) the reduction
of the oxygen molecules to hydrogen peroxide, involving two electrons,
ii) the reduction of some of the hydrogen peroxide produced in reaction
i) to hydroxide, what involves 2 additional electrons. The number of
electrons is then determined by the ratio between the H2O2 diffusion
and electrochemical reaction rates. When the electrochemical reaction
rate for H2O2 reduction is much higher than the diffusion rate, the
number of transferred electrons would be close to 4, whereas a high
mobility of H2O2 would decrease the contact time with the active sites,
and a relevant part of H2O2 could diffuse out the surface of the elec-
trode, delivering a number of electrons for ORR close to 2. Then, the
presence of micropores (which favour the adsorption of H2O2) may
increase the residence time of H2O2, increasing the probability for
further reduction. As KUA loading increases, so does the amount of
micropores and residence time of H2O2 and, consequently, the H2O2
reduction rate.
3.6. Electrochemical reduction of hydrogen peroxide in RRDE
In order to check the validity of this assumption, the electro-
chemical reduction of H2O2 was analyzed under similar conditions to
A. Gabe et al.
those in ORR measurements (Fig. 3). The concentration of hydrogen
peroxide was 3 mM so that it would be in agreement with the con-
centration of H2O2 in the micropores produced by the 2-electron re-
duction reaction of O2. Fig. 4A compares the LSV curves for all the
samples recorded in 3 mM H2O2 and N2-saturated electrolyte. A net
reduction current is observed as potential decreases, and the three
porous samples delivered a much higher reduction current than XC72.
In addition, the LSVs for hydrogen peroxide reduction share the same
potentials intervals as those observed in the transition between the 2
and the 4 electrons pathway detected in Figs. 3 and S6. The onset po-
tentials for the electrochemical reduction of H2O2 follow the order
KUA ≫ CB ∼ AC > XC72. This order is similar to that of VDRN2,
Table 1. The shape of the LSV reveals that no limiting current is
achieved (diffusion coefficient of hydrogen peroxide is
0.9·10−5 cm2 s−1, D H 2O2 is estimated to be 0.0074 cm s−1, and therefore
2·L
Levich theory predicts 4.18 mA cm−2 at 1600 rpm for 3 mM H2O2),
probably due to the poor electron transfer rate of active sites located
within micropores and the presence of an additional internal mass
transfer limitation.
When the ORR activity is determined in the presence of H2O2,
Fig. 4B–D, the resulting LSV are in agreement with a linear addition of
the ORR and hydrogen peroxide reduction reactions experiments.
Therefore, no competition between H2O2 and O2 for the active sites
seems to take place.
The electrochemical reduction of H2O2 by porous carbon electrodes
was previously studied. Most works relate this reduction with a longer
contact time between the surface of the carbon electrodes and hydrogen
peroxide [44,47,48]. It is well known the higher adsorption potential of
narrow micropores and their molecular sieve properties [38,49]. The
presence of micropores with sizes smaller than 0.7 nm was related to a
higher storage capacity for the electrochemical hydrogen storage [50],
and improved capacitance in supercapacitors [51], proving that these
pores are active in electrochemical processes. Thus, higher O2 and H2O2
concentrations due to their favorable adsorption in the narrow micro-
porosity of the most active samples (KUA and AC in a lesser degree)
could explain the improved catalytic activity.
3.7. Effect of rotation rate on ORR activity
Experiments at different rotation rates were conducted to discard
that H2O2 formed during ORR is electrochemically reduced at the ex-
ternal surface of the electrode due to an increased contact time (Fig.
S7). This premise is derived from the work by Zhou et al., who recently
proposed that ORR of nitrogen-doped carbon materials cannot be
properly analyzed using RRDE due to the dependence of hydrogen
peroxide concentration at the surface of their catalysts [52]. We have
observed that the selectivity towards hydrogen peroxide at less positive
potentials is unrelated to the rotation rate or even slightly decreases
with the rotation rate (which is the opposite behavior to that proposed
by Zhou et al.), Fig. S7. Therefore, the internal diffusion rates of oxygen
and hydrogen peroxide, which are independent of the rotation rate, are
responsible for the conversion rate of hydrogen peroxide to hydroxide
in microporous carbon electrodes.
3.8. Effect of oxygen adsorption time
Seredych et al. recently proposed that the strong adsorption of
oxygen in narrow micropores weakens the OeO bond, favoring the
oxygen reduction to water via 4-electron pathway at less positive po-
tentials [30]. These micropores are highly hydrophobic and they
withdraw oxygen from the electrolyte that is later electrochemically
reduced. Nevertheless, oxygen must diffuse through wetted pores be-
fore reaching the narrower ones. The low mass transfer rate in wetted
pores has been pointed out as the origin of the performance decay of
microporous MeeNeC catalysts; as pores get flooded with water
formed during ORR, oxygen change its diffusion mode from gaseous
457
Journal of Power Sources 412 (2019) 451–464
diffusion to liquid diffusion through the wetted pores, what results in a
higher mass transfer resistance [12]. Therefore, a very low oxygen mass
transfer rate is expectable. In this sense, we checked if ORR activity
increases due to a higher local concentration of O2 within the pores. For
that purpose, Fig. S8 compares the LSV ORR profile at 1600 rpm of KUA
recorded right after a previous LSV experiment and after 12 h of O2
bubbling. If oxygen has diffusive limitations through narrow micro-
pores, the additional adsorption time of 12 h should render a higher
reduction current due to a larger O2 concentration in the pores. Con-
trarily, the LSV profiles for both experiments are fairly similar, so either
the oxygen concentration within narrow micropores is not playing a
relevant effect, or all the active sites hosted in these micropores can be
reached by oxygen in short times. Consequently, hydrogen peroxide
reduction to water at micropores, rather than the direct oxygen re-
duction to water, is considered the prevailing mechanism behind the
improved ORR activity shown by porous carbon electrodes at less po-
sitive potentials.
3.9. Structural order and ORR activity
The higher ORR activity of microporous activated carbons has been
studied by Qu and it has been connected to their structural order
parameters derived from XRD [31]. In the case of the porous carbons
studied in this work, no relationship between order-related parameters
calculated from the XRD profiles, as those compiled in Table 1, and
ORR activity is not found (i.e., see the well-defined and intense (002)
peak of XC72 and the almost absent peak in KUA, Fig. S2, whereas ORR
activity is higher in KUA). Nonetheless, a higher concentration of edge
sites is in general expectable in microporous activated carbons, so the
identification of edge sites as an active site for ORR could be valid [2],
and, in fact, it was claimed elsewhere [9,21,47,53–55].
3.10. Modeling of ORR in porous carbon materials
The kinetic rate of oxygen reduction reaction relies on the reaction
mechanism, which is highly elusive owing to the high irreversibility of
the reaction. It has been described for platinum and other metals [56],
but still is far from being proven in the case of carbon materials. The
following set of reactions is generally accepted for explaining the hy-
drogen peroxide pathway in alkaline solutions [9,10,57]. In it, oxygen
adsorption on a free active site, f , is followed by four consecutive
surface reactions
KO 2
O2 + f (O2 ) (3)
k Or12
(O2) + e (O2 ) (4)
k Or22
(O2 ) + H2 O + e (HO2 ) + OH (5)
k Hr12 O2
(HO2 ) + H2 O + e + 2O
k H2rO
1
2 (6)
k Hr22 O2
(OH ·) + e (OH ) (7)
Both hydroperoxide and hydroxide anions can desorb from the ac-
tive site, explaining the intermediate n values between 2 and 4 e−
K Hp2 O2
(HO2 ) HO2 + f (8)
p
KOH
(OH ) OH + f (9)
Considering steady state approximation for the adsorbed superoxide
anion and hydroxyl species in Eqns. (4)–(7) and the adsorption con-
stants, Eqns. (3), (8) and (9), the * site balance can be solved. It allows
A. Gabe et al. Journal of Power Sources 412 (2019) 451–464

to define the electrochemical reaction rates, JK (mol cm−2 s−1) of Cip stands for the concentration (mol·cm−3) of the i reagent in the pores
oxygen and hydrogen peroxide supposing that the rate determining involved in the reduction reaction. Standard potentials of 1.17 V vs
steps are the formation of the superoxide anion and the formation of the NHE for ORR and 0.935 V vs NHE for H2O2 reduction reaction at pH 13
hydroperoxide anion have been considered for calculating the overpotential. Furthermore, f
stands for F/RT, where R is gas constant (8.314 J K−1 mol−1) and T is
JOK2 = k Or12 ·C ( O 2) the temperature (298 K).
(k Or12 ·C t )·KOads
2
·CO2 Once the kinetic rate expression is defined, it must be coupled with
= r ,1
kO r1
kH
2 O2 1 1 the mass transfer rates. According to our experimental data, the pre-
1+ 1+ 2
·KOads ·CO2 + 1 + · ·CHO2 + ads ·COH
r ,2
kO
2
2 r2
kH
2 O2
r 1
+ kH
2 O2 K ads
H2 O2 KOH sence of narrow micropores can have an influence in the residence time
of the reactants and reaction products through internal diffusional
(10)

r1
kH
2 O2 1
r1
(k Or12 ·C t )·KOads
2
·CO2 r1
(k H2 O2
·C t ) r 1
(k H2 O2
·C t )· r2 r 1 · ads
·CHO2
kH kH + kH KH
2 O2 2 O2 2 O2 2 O2
JHK2 O2 = k Or12 ·C (O2) + k Hr 2 O
1
2
· r2
·C (HO2 ) k Hr12 O2 ·C (HO2 ) =
kH2 O2
+ k Hr 2 O
1
2
kOr ,1 r1
kH
2 O2 1 1
1+ 1+ 2
·KOads ·CO2 + 1 + · ·CHO2 + ·COH
k O,2
r
2
2 r2
kH
2 O2
r 1
+ kH
2 O2
ads
KH
2 O2
K ads
OH (11)

The detailed mathematical derivation of the concentration of free

sites is presented in the ESI. In this study, we assume that adsorption
constants and concentration of active sites are not dependent with the
potential. In this sense, Andreas and Conway reported that pseudoca-
pacitance contribution in carbon materials is mostly relevant at low pHs
and have a reduced contribution at the pH employed in this work [58].
Note that the kinetic constants (k ir ) are potential-dependent, in
agreement to the Butler-Volmer model [33].
k ir = k 0,r i·exp (
i ·f · ) (12)
Previous studies have also reported first-order kinetics for the re-
duction of oxygen in carbon electrodes [57]. Similarly, hydrogen per-
oxide reduction experiments on AC, CB and KUA point out a linear
increase of the current for [H2O2] of 2 and 3 mM, Fig. S9. Considering
Equations (10) and (11), first order kinetics for oxygen or hydrogen
peroxide reduction is attained when the term ads ·COH prevails over
1
kOr,1 r1
kH
1+ 2
·K Oads ·CO2 + 1+ 2 2 O
r ,2
kO 2 r2
kH r 1
+ kH
2 2 O2 2 O2
likely to happen in alkaline electrolyte, where hydroxide concentration
is several orders of magnitude higher than oxygen and hydrogen per-
oxide ones. In addition, recent DFT studies about the adsorption of ORR
intermediate species in carbon materials revealed that the O2 adsorp-
tion energy on carbon is weaker than that of OH− and HO2−, whereas
the latter two species show similar energy of interaction with edge sites
[59]. In agreement with this observation, the relevance of hydroxide
concentration in the kinetic rate of ORR on carbon materials has been
confirmed by Yang and McCreery [10]. The reduction experiments
performed in presence of both O2 and H2O2 also revealed the absence of
competitive adsorption, Fig. 4, confirming that the adsorbed amount of
H2O2 does not affect the kinetic rate at the studied conditions. Thus,
since the number of parameters for fitting would be very high and the
catalysts under study are low crystallinity materials with a high het-
erogeneity and information regarding the nature and structure of the
active sites is not well known, what is an important difference com-
pared to highly crystalline materials like Pt based catalysts [56], we will
use in this study first order kinetics. This will permit us to deepen into
the role of the microporosity in these materials. Considering first order
kinetic rates, simplified expressions for the 2 + 2 reaction pathway
could be derived as follows
JiK , j = k ir , j· Cip = k 0,r , ji·exp ( i ·f ·
Where is the charge transfer rate constant (cm·s ) for the reduc-
k 0,r ,ij
tion reaction of the i reagent (either O2 or H2O2) in the narrow or wide
micropores (being j = n and j = w, respectively). Note that this para-
meter is assumed to include adsorption constants, amount of active sites
and the rest of terms included in the denominator of Eqns. (10) and
(11). i is the electron transfer coefficient, is the overpotential (V) and
limitations. We present a model to describe the ORR in porous carbons
under the following assumptions.
• Narrow and wide micropores can have different mass transfer rates,
JD, and density of active sites, C t . The latter value is numerically
included within the charge transfer rate constant, (k 0,r , ji· C t ) . Hence
k 0,r , ji will show different values for the wider and the narrow pores
during the parameter optimization. Oxygen and hydrogen peroxide
concentration will also differ for narrow and wide micropores.
• Narrow and wide micropores are supposed to be randomly dis-
tributed in carbon particles, so that the entrances of the pores are
equally accessible to oxygen, ruling out the prerequisite of oxygen
diffusing through wider pores before reaching the narrower ones.
However, they will differ in j , the parameter representing the ratio
between the effective surface area for mass transfer of narrow
KOH
(j = n) or wide (j = w) micropores with respect to the geometrical
·
1
· CHO2 . This scenario is surface area of the disk.
•
ads
KH
O
2 2 Oxygen and hydrogen peroxide diffusional rates ( JOD2, jp and JHD2, jp O2 ,
mol cm−2 s−1, where j stands for narrow, n, or wide, w, micropores)
are estimated by applying Fick's law along the perpendicular di-
rection to the electrode surface (defined as x direction):
COjp2
JOD2, jp = j · D O2 ·
x (14)
CHjp2 O2
JHD2, jp
O2 = j · D H2 O2 ·
x (15)
In them, DO2 and D H2 O2 are the diffusivity of oxygen and hydrogen
peroxide. Terms accounting for mobile ionic species and Ohm's law
are not included, given that oxygen is a nonpolar molecule, and that
conductivity of the electrolyte is high enough and unaffected by the
concentration of the reacting species, since a large excess of sup-
porting electrolyte is present. As previously mentioned pseudoca-
pacitance in alkaline electrolyte is low and is not expected to affect
oxygen mobility in the porosity of carbon particles. A finite differ-
ence approximation for the oxygen and hydrogen peroxide con-
centration is applied between x = 0 and x = L (L being the electrode
thickness) in order to solve Eqns. (14) and (15), using similar ap-
proach to Refs. [33,56].
)·Cip (13) • D
The O2 mass transfer constant is set to match the diffusion limited
2·L
−1 mass transfer constant for oxygen derived from the RDE theory [56],
so that it is predicted in accordance to Eqn. (2), having a value of
0.0125 cm s−1 for 1600 rpm. Additional mass transfer limitations at
macropore, mesopore or micropore level on the electrode are not
considered for oxygen diffusion, since no additional mass transfer
limitations were found for ORR, Fig. S8. Absence of mass transfer in

458
A. Gabe et al. Journal of Power Sources 412 (2019) 451–464

the porosity of the particles can be understood as a consequence of
the different size of the diffusion layers; in a similar way to that
exposed by Bard about diffusion on the surface of rough electrodes
[33]. Thus, Eqn. (14) allows to define the oxygen mass transfer rates
in narrow microporosity and wide microporosity, JOD2, n and JOD2, w ,
respectively.
• Hydrogen peroxide in wide micropores follows the same diffusion
mechanism as oxygen, with H 2O2 having a value of 0.0074 cm s−1 at
D
2·L
1600 rpm in accordance to the RDE theory. Given the large volume
of the cell, the low formation rate of hydrogen peroxide and the
short analysis times, the bulk concentration of hydrogen peroxide,
CHb 2 O2 , can be considered negligible. Differently, hydrogen peroxide
reduction has shown additional mass transfer limitations in carbon
materials, Fig. 4, so that an effective diffusion coefficient for H2O2 in
D np
narrow pores, H2 O2 , needs to be defined and is selected as a model
2·L
parameter.
activity of porous carbon materials in the RRDE system. Differently, the
model that considers homogeneous ORR activity in micropores pro-
duces a largest source of divergence in the number of electrons trans-
ferred at potentials higher than 0.5 V. This model cannot explain the
slope changes in the number of transferred electrons (see Figs. S10A
and B), making necessary the consideration of two H2O2 reaction rates
of different activity, like in the narrow/wide micropore model. In a
similar way, the relevance of the backward reaction of Eqn. (6) was
checked by solving the model setting k H2rO, n2 to zero and comparing the
optimized solutions, Figs. S10C and D. The small impact on the fitting
moved us to remove the reaction from the model for the sake of
D np
clearness. Finally, a linear relationship was found between H2 O2 and
2·L
k 0,r ,Hnp2 O2 what means that the same fitting can be reached by keeping
constant the ratio between these parameters. Therefore, the optimiza-
tion parameters were cut down to eight, with a new model parameter
np
DH

•
being defined as R H2 O2 = k 0,r ,Hnp2 O2 / 2·2L 2 .
O
The current registered at the disk ( jDORR , mA cm−2) is produced by
Tafel analyses were performed for the LSV curves in the kinetically
the electrochemical reduction of oxygen to hydrogen peroxide (2
controlled region to calculate the kinetic parameters for the ORR to
e−, jO2 ) and the electrochemical reduction of hydrogen peroxide to
hydrogen peroxide. These constants were calculated from Eqns. (13)
hydroxide (2 e−, jH2 O2 ).
and (18) assuming that the oxygen concentration is constant in the
jDORR = jO2 + jH2 O2 (16) pores, COp2 , and is the same as in the bulk of the solution. Tafel plots are
shown in Fig. S10A. O2 is estimated from the slope of the linear region,

• The effective number of electrons transferred (n) can be estimated being 0.82, 0.78, 0.86 and 0.89 for KUA, AC, CB and, XC72 respec-
tively, while k 0,r O2 values of 10·10−9, 7.1·10−9, 1.4·10−9 and
from the ratio between the reduction currents of oxygen and hy-
0.25·10−9 cm s−1 are determined from the Y-intercept for the same set
drogen peroxide:
of samples. The kinetic parameters obtained from Tafel analyses and
jH2 O2 the least square fitting of the proposed model are comparable. Tafel
n = 2· 1 + slope of KUA, AC, CB and, XC72 are similar (around
jO2 (17)
45–50 mV decade−1) to those reported in previous investigations for
carbon materials [28].
• The electrochemical reaction rates are those defined in Eqns. (10) The sum of the parameters w + N (which is proportional to the
and (11) for the full kinetic mechanism, and those from Eqn. (13) for effective surface area) follows the order XC72 > AC > CB > KUA.
the simplified kinetic mechanism. No clear relationship can be made with textural parameters derived
• The intrinsic electrochemical reaction rates for O2 and H2O2 re-
duction (respectively denoted as JOK2 and JHK2 O2 , mol cm−2 s−1) are
from the N2 adsorption isotherms. This can be explained by the dif-
ferent sizes of the micro and mesoporosity of the samples, that are
related to the disk current through the number of electrons trans- defined in the range of a few to several nanometers, and that of the
ferred (2 e−) and the Faraday constant, F: diffusion layer, which is expected to show a thickness in the range of
jDORR = jO2 + jH2 O2 = 2· F · (JOK2, n + JOK2, w + JHK2,O several micrometers. Therefore, on the scale of the diffusion layer, the
2)
n
+ JHK2, O
w
(18)
electrode can be considered as flat [33]. Effective surface areas lower
2

than 1 can reflect that the surface of the disk is not fully covered for
•O 2 and H2O2 concentrations in narrow and wide micropores are
estimated from their mass balances assuming that quasi-equilibrium
conditions are achieved (i.e. mass transfer rates are equal to the
kinetic rates)
• The model parameters are those defining the intrinsic reaction
constants, i.e. k 0,r ,Ow2 , k 0,r ,Hw2 O2 , k 0,r ,On 2 , k 0,r ,Hn2 O2 , k 0, H
r,n
2 O2 O2
and H2 O2 , and
np
DH
those related with mass transfer rates, i.e. w , n and 2·2L 2 . These
O

parameters are optimized to minimize the square difference be-

tween the experimentally measured current densities and number of
electrons and those predicted by the model (Eqns. (17) and (18)).
• Additionally, a model where homogeneous activity on narrow and
wide micropores is assumed was also tested.

Fig. 5 presents a graphical summary of both proposed models for

estimating the ORR activity during RRDE experiments of porous carbon
materials.

3.11. Application of the proposed model for predicting the ORR activity of
carbon materials

The ORR model described in the previous section was fitted to the
LSV profiles at 1600 rpm of the porous carbon samples shown in Fig. 3.
Fig. 6 compares the experimental data with the obtained theoretical
curves, while Table 3 compiles the optimized model parameters. Fig. 6
demonstrates that, in general, the model is able to reproduce the ORR
Fig. 5. Schemes for the heterogeneous micropores (upper) and homogeneous
micropores (bottom) models proposed for the mathematical description of the
ORR mechanism of porous carbon materials.

459
A. Gabe et al. Journal of Power Sources 412 (2019) 451–464

Fig. 6. Experimental (dots) and modeled (lines) for A) LSV and B) number of transferred electrons profiles during ORR at 1600 rpm in 0.1 M KOH. KUA (blue), AC
(red), CB (black) and XC72 (green). Scan rate: 5 mV s−1. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version
of this article.)

Table 3 have higher catalytic activity probably due to the differences in mass
Optimized model parameters. transfer between both type of pores that produce higher residence time
Parameter KUA AC CB XC72
for H2O2 within the narrow micropores.
The LSV curve of KUA for 1.00 mg cm−2 was also been fitted to the
w 0.40 0.50 0.69 0.80 proposed model, Fig. S11B. The effective surface area parameters
n 0.35 0.41 0.14 0.14 should account for the increase in the catalyst loading, owing to a larger
O2 0.83 0.79 0.83 0.87 surface area available for diffusion. The charge transfer rates could also
k 0,r ,Owp (cm s−1) 1.0·10−8 3.3·10−9 1.6·10−9 2.0·10−9
2 increase, since they include density of active sites along with other
k 0,r ,Onp2 (cm s−1) 3.4·10−9 1.1·10−9 4.7·10−10 3.2·10−10 kinetic and adsorption constants, as was discussed in the definition of
H2 O2 0.22 0.21 0.25 0.21 Eqn. (13). Thus, the model was first applied fixing all parameters
R H2 O2 1.6·10−2 1.2·10−3 3.1·10−3 6.6·10−3 compiled in Table 3 for KUA and the n to w ratio to 0.837 (i.e. the
k 0,r ,Hwp (cm s−1) 1.7·10−6 1.3·10−6 1.2·10−6 9.5·10−7
2 O2 value observed in the model fitting in Table 3 for KUA). Fixing this ratio
tries to represent the independency of the textural properties of the
catalysts with respect to the carbon loading. Although the only fitting
some of the samples. A similar argument was provided for RRDE in- parameter was w , which gets a value of 0.55, the model successfully
volving electrocatalytic nanoparticles [43,46]. The effective surface describes the experimental LSV curve, black curve on Fig. S11B. The
area of the electrodes can be also affected by the amount of Nafion® in effective surface area is now equal to one, what is explained by the
the thin film or the particle distribution on the electrode surface [60]. surface the electrode fully covered by the carbon sample. An additional
Nevertheless, samples XC72, with the lowest micropore volume, and fitting was also performed fixing electron transfer coefficients, while
KUA, with the highest micropore volume, have the highest and the the charge transfer rates for wide micropores are optimized. Those of
lowest parameter, respectively. This is in agreement with the ex- narrow pores are obtained from the wide pore ones by using the narrow
pected higher mass transfer limitations in highly microporous mate- to wide pore parameter ratios from Table 3. The new fitting results in
rials. no changes in the oxygen charge transfer rates, which seems to be in-
However, when the ratio N /( N + W ) is compared to equivalent dependent of loading. However, the hydrogen peroxide charge transfer
ratios in the textural characterization, the best relationship is found in wide pores increases 2.66 times, while RH2O2 for narrow pores in-
with V DR
CO2 N2
/(VDR + Vmes ) (R2 = 0.988, Fig. 7A). The latter parameter can creases 1.28 times, indicating that more active sites for H2O2 reduction
be described as the ratio between narrow micropores and the total pore are available as the loading increases. Considering that the charge
volume. Interestingly, the charge transfer rate constants, k 0,r ,Owp2 and k 0,r ,Onp2 , transfer rate constant includes the adsorption constant, a possible ex-
seem to be linearly connected to the N2 and CO2 micropore volumes of planation for the differences observed for O2 and H2O2 kinetic constant
the samples, respectively (Fig. 7B and C). However, the value of me- rates could be that adsorption of oxygen is more favorable than ad-
sopore volume is unconnected, confirming that the higher amount ac- sorption of H2O2. Then, oxygen adsorption could be close to saturation
tives sites are present in micropores. for both catalytic loadings, having a minor impact on the charge
The H2 O2 electron transfer coefficient values are low compared to transfer rate. Differently, the amount of adsorbed H2O2 could increase
O2 (Table 3). A similar value was recently found for the electrocatalytic linearly with the catalyst loading and produce larger changes in the
reduction of hydrogen peroxide on dispersed gold nanoparticles [61]. respective charge transfer rate. Accordingly, further experiments and
The low activity of undoped carbon materials towards H2O2 electro- more complex description of the electrode system is required for pro-
chemical reduction is well-known, and it is beneficial for the prepara- viding a more accurate description of the effect of the electrode loading.
tion of electrodes selective to the formation of hydrogen peroxide [48]. Once the model was verified, it can be also useful as a tool to
In the case of R H2 O2 , the best correlation is achieved for VDRCO2/VMES, analyze the effect of each parameter on the ORR activity of porous
Fig. 7D. Although linear trend between VDRN2 and k 0,r ,Hwp2 O2 has been
carbon materials. In this sense, some of the model parameters that
obtained (not included), the correlation is poorer than for the rest of
describe KUA activity were varied one by one and the outcomes are
analyzed parameters (R2 = 0.757). Interestingly, the ratio
plotted in Fig. 8. In first place, both N and W were increased 20%. The
k 0,r ,Hwp2 O2 / ( D H2 O2
2·L ) is one order of magnitude lower than that of narrow LSV current on the diffusive controlled region has accordingly in-
micropores, R H2 O2 , for all samples, meaning that narrow micropores creased, Fig. 8A. However, a negligible impact on the number of

460
A. Gabe et al.
Fig. 7. Relationships between textural and model parameters A) effective surface area ratios and narrow micropore volume. B) wide micropore charge transfer rate
constant and micropore volume. C) narrow micropore charge transfer rate constant and narrow micropore volume. D) H2O2 charge transfer to mass transfer rate ratio
and narrow micropore volume.
transferred electrons is observed (i.e. increases from 3.02 to 3.04 at
0.2 V, Fig. 8B). The effect of shifting the pore size distribution to nar-
rower sizes was also explored, for what the sum of N and W has been
fixed, while the ratio of N /( N + W ) was increased from 0.45 to 0.60.
The impact on the LSV current is low but, the number of electrons
transferred increases ca. 10%, showing a value of 3.31 at 0.2 V.
The effect of the charge transfer rate constants for oxygen reduction
on wide and narrow pores is evaluated on Fig. 8C. 50% increase in the
value of charge transfer rate constants for either wide or narrow mi-
cropores seems to have a small impact on the shape of the resulting LSV
curve. However, detailed analyses of the kinetically controlled region,
Fig. 8C, revealed an increase of 6 and 11 mV on the onset potential
(j = −0.1 mA cm−2) for narrow and wide micropores, respectively. No
changes in the number of electron transferred are observed, since it is
governed by the hydrogen peroxide reduction reaction.
The effect of model parameters related to hydrogen peroxide were
also analyzed. The value of charge transfer rate of H2O2 reduction on
wide pores and RH2O2 of narrow pores were 3-fold increased. The spe-
cific current on the LSV and the number of electrons increase at low
potentials in the latter case and at medium potentials in the former,
Fig. 8D and E.
Lastly, the effect of increasing 5% the electron transfer coefficient
for O2 and H2O2 reactions was tested. Small increase of αO2 has a strong
effect on the onset potential (positive shifting of 16 mV), while αH2O2
marginally enhances the number of electron transferred at medium and
low potentials (higher currents on the diffusional controlled region on
Fig. 8F). However, this work has not found any clear relationship be-
tween electron transfer coefficients and porosity.
461
Journal of Power Sources 412 (2019) 451–464
4. Conclusions
The activity of porous carbon materials in the ORR in alkaline
medium, and in absence of any additional catalyst, is undoubtedly re-
lated to their porosity. By characterizing carbon materials with dif-
ferent pore size distribution and pore structure, we have been able to
demonstrate that the ORR is accurately described by a two-wave pro-
cess, where oxygen is reduced to hydrogen peroxide in a first step at
intermediate potentials, while hydrogen peroxide is subsequently re-
duced to hydroxide at low potentials. The combination of both reac-
tions rises the number of transferred electrons to intermediate values
between 2 and 4.
The onset potentials for both reactions seems to be related to the
presence of a well-developed microporosity. ORR experiments in RDE
have confirmed that microporosity is certainly correlated to a high
activity in this reaction, and that the presence of H2O2 does not inter-
fere in the rate of the ORR. In addition, ORR experiments performed
with different loadings demonstrated that the amount of sample has a
critical impact on the ORR activity of highly microporous materials.
A mathematical model that describes the reaction rate and number
of electrons transferred during ORR is derived. It takes into account the
O2 and H2O2 mass transfer rate, two consecutive reduction reactions
and different activity of narrow and wide micropores. The model gives
a successful description of the ORR activity of porous carbon electrodes
provided that the two types of micropores are considered, verifying the
claims about the importance of microporosity on the reduction of di-
oxygen and of narrow microporosity on the reduction of hydrogen
peroxide. The differences in adsorption potential and mass transfer
A. Gabe et al.
Fig. 8. Effect of model parameters on the ORR activity of KUA. A-B) Effect of i on the LSV and number of electron transferred. C) Effect of k 0,r ,Oi 2 on the simulated
LSV. D, E) Effect of hydrogen peroxide kinetic parameters. F) Effect of electron transfer coefficients.
between both types of pores may explain the higher activity for H2O2
reduction observed for narrow micropores. This work can be a guide-
line for the proper analysis of ORR activity of new catalysts based on
carbon materials, where the effect of porosity is frequently omitted.
Acknowledgment
This work was supported by MINECO (CTQ2015-66080-R MINECO/
FEDER) and Heiwa Nakajima Foundation.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.jpowsour.2018.11.075.
462
Journal of Power Sources 412 (2019) 451–464
References
[1] P. Trogadas, T.F. Fuller, P. Strasser, Carbon as catalyst and support for electro-
chemical energy conversion, Carbon 75 (2014) 5–42, https://doi.org/10.1016/j.
carbon.2014.04.005.
[2] L.R. Radovic, Surface chemical and electrochemical properties of carbons, in:
F. Beguin, E. Frackowiak (Eds.), Carbons for Electrochemical Energy Storage and
Conversion Systems, Taylor & Francis (CRC Press), Boca Raton, FL, 2010, pp.
163–219.
[3] A.L. Dicks, The role of carbon in fuel cells, J. Power Sources 156 (2006) 128–141,
https://doi.org/10.1016/j.jpowsour.2006.02.054.
[4] F. Jaouen, E. Proietti, M. Lefèvre, R. Chenitz, J.-P. Dodelet, G. Wu, H.T. Chung,
C.M. Johnston, P. Zelenay, Recent advances in non-precious metal catalysis for
oxygen-reduction reaction in polymer electrolyte fuel cells, Energy Environ. Sci. 4
(2010) 114–130, https://doi.org/10.1039/C0EE00011F.
[5] L. Dai, Y. Xue, L. Qu, H.-J. Choi, J.-B. Baek, Metal-free catalysts for oxygen re-
duction reaction, Chem. Rev. 115 (2015) 4823–4892, https://doi.org/10.1021/
cr5003563.
[6] A. Morozan, B. Jousselme, S. Palacin, Low-platinum and platinum-free catalysts for
the oxygen reduction reaction at fuel cell cathodes, Energy Environ. Sci. 4 (2011)
1238–1254, https://doi.org/10.1039/C0EE00601G.
A. Gabe et al.
[7] Y. Nie, L. Li, Z. Wei, Recent advancements in Pt and Pt-free catalysts for oxygen
reduction reaction, Chem. Soc. Rev. 44 (2015) 2168–2201, https://doi.org/10.
1039/C4CS00484A.
[8] C. Tamain, S.D. Poynton, R.C.T. Slade, B. Carroll, J.R. Varcoe, Development of
cathode architectures customized for H2/O2 metal-cation-free alkaline membrane
fuel cells, J. Phys. Chem. C 111 (2007) 18423–18430, https://doi.org/10.1021/
jp076740c.
[9] M. Gara, R.G. Compton, Activity of carbon electrodes towards oxygen reduction in
acid: a comparative study, New J. Chem. 35 (2011) 2647, https://doi.org/10.1039/
c1nj20612e.
[10] H.-H. Yang, R.L. McCreery, Elucidation of the mechanism of dioxygen reduction on
metal‐free carbon electrodes, J. Electrochem. Soc. 147 (2000) 3420–3428, https://
doi.org/10.1149/1.1393915.
[11] T. Ikeda, M. Boero, S.F. Huang, K. Terakura, M. Oshima, J. Ozaki, Carbon alloy
catalysts: active sites for oxygen reduction reaction, J. Phys. Chem. C 112 (2008)
14706–14709, https://doi.org/10.1021/jp806084d.
[12] M. Shao, Q. Chang, J.-P. Dodelet, R. Chenitz, Recent advances in electrocatalysts for
oxygen reduction reaction, Chem. Rev. 116 (2016) 3594–3657, https://doi.org/10.
1021/acs.chemrev.5b00462.
[13] J. Quílez-Bermejo, C. González-Gaitán, E. Morallón, D. Cazorla-Amorós, Effect of
carbonization conditions of polyaniline on its catalytic activity towards ORR. Some
insights about the nature of the active sites, Carbon 119 (2017) 62–71, https://doi.
org/10.1016/j.carbon.2017.04.015.
[14] A. Gabe, J. García-Aguilar, Á. Berenguer-Murcia, E. Morallón, D. Cazorla-Amorós,
Key factors improving oxygen reduction reaction activity in cobalt nanoparticles
modified carbon nanotubes, Appl. Catal. B Environ. 217 (2017) 303–312, https://
doi.org/10.1016/j.apcatb.2017.05.096.
[15] T. Sharifi, G. Hu, X. Jia, T. Wågberg, Formation of active sites for oxygen reduction
reactions by transformation of nitrogen functionalities in nitrogen-doped carbon
nanotubes, ACS Nano 6 (2012) 8904–8912, https://doi.org/10.1021/nn302906r.
[16] G.-L. Chai, Z. Hou, D.-J. Shu, T. Ikeda, K. Terakura, Active sites and mechanisms for
oxygen reduction reaction on nitrogen-doped carbon alloy catalysts: stone–wales
defect and curvature effect, J. Am. Chem. Soc. 136 (2014) 13629–13640, https://
doi.org/10.1021/ja502646c.
[17] C.V. Rao, C.R. Cabrera, Y. Ishikawa, In search of the active site in nitrogen-doped
carbon nanotube electrodes for the oxygen reduction reaction, J. Phys. Chem. Lett.
1 (2010) 2622–2627, https://doi.org/10.1021/jz100971v.
[18] L. Zhang, Z. Xia, Mechanisms of oxygen reduction reaction on nitrogen-doped
graphene for fuel cells, J. Phys. Chem. C 115 (2011) 11170–11176, https://doi.org/
10.1021/jp201991j.
[19] R. Chen, H. Li, D. Chu, G. Wang, Unraveling oxygen reduction reaction mechanisms
on carbon-supported Fe-phthalocyanine and Co-phthalocyanine catalysts in alka-
line solutions, J. Phys. Chem. C 113 (2009) 20689–20697, https://doi.org/10.
1021/jp906408y.
[20] U. Tylus, Q. Jia, K. Strickland, N. Ramaswamy, A. Serov, P. Atanassov, S. Mukerjee,
Elucidating oxygen reduction active sites in pyrolyzed metal–nitrogen coordinated
non-precious-metal electrocatalyst systems, J. Phys. Chem. C 118 (2014)
8999–9008, https://doi.org/10.1021/jp500781v.
[21] K. Waki, R.A. Wong, H.S. Oktaviano, T. Fujio, T. Nagai, K. Kimoto, K. Yamada, Non-
nitrogen doped and non-metal oxygen reduction electrocatalysts based on carbon
nanotubes: mechanism and origin of ORR activity, Energy Environ. Sci. 7 (2014)
1950–1958, https://doi.org/10.1039/C3EE43743D.
[22] C. Domínguez, F.J. Pérez-Alonso, M. Abdel Salam, S.A. Al-Thabaiti, A.Y. Obaid,
A.A. Alshehri, J.L. Gómez de la Fuente, J.L.G. Fierro, S. Rojas, On the relationship
between N content, textural properties and catalytic performance for the oxygen
reduction reaction of N/CNT, Appl. Catal. B Environ. 162 (2015) 420–429, https://
doi.org/10.1016/j.apcatb.2014.07.002.
[23] G.A. Ferrero, K. Preuss, A.B. Fuertes, M. Sevilla, M.-M. Titirici, The influence of pore
size distribution on the oxygen reduction reaction performance in nitrogen doped
carbon microspheres, J. Mater. Chem. 4 (2016) 2581–2589, https://doi.org/10.
1039/C5TA10063A.
[24] J. Park, Y. Nabae, T. Hayakawa, M. Kakimoto, Highly selective two-electron oxygen
reduction catalyzed by mesoporous nitrogen-doped carbon, ACS Catal. 4 (2014)
3749–3754, https://doi.org/10.1021/cs5008206.
[25] J. Maruyama, K.I. Sumino, M. Kawaguchi, I. Abe, Influence of activated carbon pore
structure on oxygen reduction at catalyst layers supported on rotating disk elec-
trodes, Carbon 42 (2004) 3115–3121, https://doi.org/10.1016/j.carbon.2004.07.
023.
[26] F. Jaouen, M. Lefèvre, J.-P. Dodelet, M. Cai, Heat-treated Fe/N/C catalysts for O2
Electroreduction: are active sites hosted in micropores? J. Phys. Chem. B 110
(2006) 5553–5558, https://doi.org/10.1021/jp057135h.
[27] M. Lefèvre, E. Proietti, F. Jaouen, J.-P. Dodelet, Iron-based catalysts with improved
oxygen reduction activity in polymer electrolyte fuel cells, Science 324 (2009)
71–74, https://doi.org/10.1126/science.1170051.
[28] A.J. Appleby, J. Marie, Kinetics of oxygen reduction on carbon materials in alkaline
solution, Electrochim. Acta 24 (1979) 195–202, https://doi.org/10.1016/0013-
4686(79)80024-9.
[29] Y. Liu, K. Li, B. Ge, L. Pu, Z. Liu, Influence of micropore and mesoporous in acti-
vated carbon air-cathode catalysts on oxygen reduction reaction in microbial fuel
cells, Electrochim. Acta 214 (2016) 110–118, https://doi.org/10.1016/j.electacta.
2016.08.034.
[30] M. Seredych, A. Szczurek, V. Fierro, A. Celzard, T.J. Bandosz, Electrochemical re-
duction of oxygen on hydrophobic ultramicroporous PolyHIPE carbon, ACS Catal. 6
(2016) 5618–5628, https://doi.org/10.1021/acscatal.6b01497.
[31] D. Qu, Investigation of oxygen reduction on activated carbon electrodes in alkaline
solution, Carbon 45 (2007) 1296–1301, https://doi.org/10.1016/j.carbon.2007.01.
463
Journal of Power Sources 412 (2019) 451–464
013.
[32] M.A. Lillo-Ródenas, D. Lozano-Castelló, D. Cazorla-Amorós, A. Linares-Solano,
Preparation of activated carbons from Spanish anthracite - II. Activation by NaOH,
Carbon 39 (2001) 751–759, https://doi.org/10.1016/S0008-6223(00)00186-X.
[33] A.J. Bard, L.R. Faulkner, Electrochemical Methods: Fundamentals and Applications,
2a, Wiley, 2001.
[34] M. Thommes, K. Kaneko, A.V. Neimark, J.P. Olivier, F. Rodriguez-Reinoso,
J. Rouquerol, K.S.W. Sing, Physisorption of gases, with special reference to the
evaluation of surface area and pore size distribution (IUPAC Technical Report),
Pure Appl. Chem. 87 (2015) 1051–1069, https://doi.org/10.1515/pac-2014-1117.
[35] M.A. Lillo-Ródenas, D. Cazorla-Amorós, A. Linares-Solano, Understanding chemical
reactions between carbons and NaOH and KOH: an insight into the chemical acti-
vation mechanism, Carbon 41 (2003) 267–275, https://doi.org/10.1016/S0008-
6223(02)00279-8.
[36] F. Zaragoza-Martín, D. Sopeña-Escario, E. Morallón, C.S.-M. de Lecea, Pt/carbon
nanofibers electrocatalysts for fuel cells, J. Power Sources 171 (2007) 302–309,
https://doi.org/10.1016/j.jpowsour.2007.06.078.
[37] D. Lozano-Castelló, D. Cazorla-Amorós, A. Linares-Solano, Usefulness of CO2 ad-
sorption at 273 K for the characterization of porous carbons, Carbon 42 (2004)
1233–1242, https://doi.org/10.1016/j.carbon.2004.01.037.
[38] D. Cazorla-Amorós, J. Alcañiz-Monge, M.A. de la Casa-Lillo, A. Linares-Solano, CO2
as an adsorptive to characterize carbon molecular sieves and activated carbons,
Langmuir 14 (1998) 4589–4596, https://doi.org/10.1021/la980198p.
[39] J. Jagiello, J.P. Olivier, 2D-NLDFT adsorption models for carbon slit-shaped pores
with surface energetical heterogeneity and geometrical corrugation, Carbon 55
(2013) 70–80.
[40] M.A. Short, P.L. Walker, Measurement of interlayer spacings and crystal sizes in
turbostratic carbons, Carbon 1 (1963) 3–9 https://doi.org/10.1016/0008-6223(63)
90003-4.
[41] K.J.J. Mayrhofer, D. Strmcnik, B.B. Blizanac, V. Stamenkovic, M. Arenz,
N.M. Markovic, Measurement of oxygen reduction activities via the rotating disc
electrode method: from Pt model surfaces to carbon-supported high surface area
catalysts, Electrochim. Acta 53 (2008) 3181–3188, https://doi.org/10.1016/j.
electacta.2007.11.057.
[42] U.A. Paulus, T.J. Schmidt, H.A. Gasteiger, R.J. Behm, Oxygen reduction on a high-
surface area Pt/Vulcan carbon catalyst: a thin-film rotating ring-disk electrode
study, J. Electroanal. Chem. 495 (2001) 134–145, https://doi.org/10.1016/S0022-
0728(00)00407-1.
[43] J. Masa, C. Batchelor-McAuley, W. Schuhmann, R.G. Compton, Koutecky-Levich
analysis applied to nanoparticle modified rotating disk electrodes: electrocatalysis
or misinterpretation, Nano Res. 7 (2014) 71–78, https://doi.org/10.1007/s12274-
013-0372-0.
[44] A. Bonakdarpour, M. Lefevre, R. Yang, F. Jaouen, T. Dahn, J.-P. Dodelet, J.R. Dahn,
Impact of loading in RRDE experiments on Fe–N–C catalysts: two- or four-electron
oxygen reduction? Electrochem. Solid State Lett. 11 (2008) B105, https://doi.org/
10.1149/1.2904768.
[45] M. Gara, K.R. Ward, R.G. Compton, Nanomaterial modified electrodes: evaluating
oxygen reduction catalysts, Nanoscale 5 (2013) 7304–7311, https://doi.org/10.
1039/c3nr01940c.
[46] K.R. Ward, M. Gara, N.S. Lawrence, R.S. Hartshorne, R.G. Compton, Nanoparticle
modified electrodes can show an apparent increase in electrode kinetics due solely
to altered surface geometry: the effective electrochemical rate constant for non-flat
and non-uniform electrode surfaces, J. Electroanal. Chem. 695 (2013) 1–9, https://
doi.org/10.1016/j.jelechem.2013.02.012.
[47] N.P. Subramanian, X. Li, V. Nallathambi, S.P. Kumaraguru, H. Colon-Mercado,
G. Wu, J.W. Lee, B.N. Popov, Nitrogen-modified carbon-based catalysts for oxygen
reduction reaction in polymer electrolyte membrane fuel cells, J. Power Sources
188 (2009) 38–44, https://doi.org/10.1016/j.jpowsour.2008.11.087.
[48] G. Coria, T. Pérez, I. Sirés, J.L. Nava, Mass transport studies during dissolved
oxygen reduction to hydrogen peroxide in a filter-press electrolyzer using graphite
felt, reticulated vitreous carbon and boron-doped diamond as cathodes, J.
Electroanal. Chem. 757 (2015) 225–229, https://doi.org/10.1016/j.jelechem.2015.
09.031.
[49] M.A. de la Casa-Lillo, J. Alcañiz-Monge, E. Raymundo-Piñero, D. Cazorla-Amorós,
A. Linares-Solano, Molecular sieve properties of general-purpose carbon fibres,
Carbon 36 (1998) 1353–1360, https://doi.org/10.1016/S0008-6223(98)00120-1.
[50] F. Béguin, K. Kierzek, M. Friebe, A. Jankowska, J. Machnikowski, K. Jurewicz,
E. Frackowiak, Effect of various porous nanotextures on the reversible electro-
chemical sorption of hydrogen in activated carbons, Electrochim. Acta 51 (2006)
2161–2167, https://doi.org/10.1016/j.electacta.2005.03.086.
[51] J. Chmiola, G. Yushin, Y. Gogotsi, C. Portet, P. Simon, P.L. Taberna, Anomalous
increase in carbon capacitance at pore sizes less than 1 nanometer, Science 313
(2006) 1760–1763, https://doi.org/10.1126/science.1132195.
[52] R. Zhou, Y. Zheng, M. Jaroniec, S.-Z. Qiao, Determination of the electron transfer
number for the oxygen reduction reaction: from theory to experiment, ACS Catal. 6
(2016) 4720–4728, https://doi.org/10.1021/acscatal.6b01581.
[53] P.H. Matter, U.S. Ozkan, Non-metal catalysts for dioxygen reduction in an acidic
electrolyte, Catal. Lett. 109 (2006) 115–123, https://doi.org/10.1007/s10562-006-
0067-1.
[54] P.H. Matter, L. Zhang, U.S. Ozkan, The role of nanostructure in nitrogen-containing
carbon catalysts for the oxygen reduction reaction, J. Catal. 239 (2006) 83–96,
https://doi.org/10.1016/j.jcat.2006.01.022.
[55] X. Chu, K. Kinoshita, Surface modification of carbons for enhanced electrochemical
activity, Mater. Sci. Eng. B 49 (1997) 53–60, https://doi.org/10.1016/S0921-
5107(97)00100-1.
[56] H.A. Hansen, V. Viswanathan, J.K. Nørskov, Unifying Kinetic and Thermodynamic
A. Gabe et al.
Analysis of 2 e– and 4 e– Reduction of Oxygen on Metal Surfaces, J. Phys. Chem. C
118 (2014) 6706–6718, https://doi.org/10.1021/jp4100608.
[57] E. Yeager, Dioxygen electrocatalysis: mechanisms in relation to catalyst structure, J.
Mol. Catal. 38 (1986) 5–25, https://doi.org/10.1016/0304-5102(86)87045-6.
[58] H.A. Andreas, B.E. Conway, Examination of the double-layer capacitance of an high
specific-area C-cloth electrode as titrated from acidic to alkaline pHs, Electrochim.
Acta 51 (2006) 6510–6520, https://doi.org/10.1016/j.electacta.2006.04.045.
[59] H. Kobayashi, N. Tomoya, S. Miyazaki, T. Miura, N. Takeuchi, T. Yamabe, DFT
study of oxygen reduction reaction on N-substituted carbon electrodes. Adsorption,
464
Journal of Power Sources 412 (2019) 451–464
IOP Conf. Ser.: Mater. Sci. Eng. 18 (2011) 122010, https://doi.org/10.1088/1757-
899X/18/12/122010.
[60] Y. Garsany, I.L. Singer, K.E. Swider-Lyons, Impact of film drying procedures on RDE
characterization of Pt/VC electrocatalysts, J. Electroanal. Chem. 662 (2011)
396–406, https://doi.org/10.1016/j.jelechem.2011.09.016.
[61] J.S. Jirkovský, M. Halasa, D.J. Schiffrin, Kinetics of electrocatalytic reduction of
oxygen and hydrogen peroxide on dispersed gold nanoparticles, Phys. Chem. Chem.
Phys. 12 (2010) 8042–8053, https://doi.org/10.1039/C002416C.